Think of it as music sampling but with the help of AI. Developers from Japan-based startup Amadeus Code recently unveiled a new app that uses deep learning to transform your favorite songs into new compositions. The app, which is available on the iTunes store, can create new songs in a matter of seconds.
Think of a hit song, search for it on the app, and within seconds you can modify the melody, chords, and tempo the algorithm creates. From there, a user can download the source files and modify the songs in their preferred midi editor.
“This is not a copy-and-paste operation for hit making,” explained the co-founder of the Tokyo-based tech startup Taishi Fukuyama. “It’s about using algorithms to suggest unfamiliar melodies and expanding one’s imagination efficiently.”
Using NVIDIA GeForce GTX 1080 GPUs, with a custom cuDNN-accelerated deep learning framework, the team trained their artificial neural network on approximately 60 years’ worth of hit songs. The team says they are continuing to enhance their algorithm by re-training it on new music as its released.
“AI has this peculiar ability to find novel solutions–some successful, some not so much,” Fukuyama said. “These are suggestions which a composer can take or leave.
Read more
AI-Generated Summary
- Amadeus Code, a Tokyo-based startup, released an iTunes app that uses deep learning to transform existing songs into new compositions within seconds.
- The team trained their artificial neural network on approximately 60 years of hit songs using NVIDIA GeForce GTX 1080 GPUs with a custom cuDNN-accelerated framework.
- Co-founder Taishi Fukuyama describes the system as a tool for suggesting unfamiliar melodies to expand a composer's imagination rather than copying existing hits.
Next Steps
- Read the Amadeus Code blog for additional details on the technology.
Powered by NVIDIA Nemotron. AI-generated content may summarize information incompletely. Verify important information. Learn more
About the Authors
